1 /*
2 ********************************************************************************
3 * Copyright (C) 1996-2015, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 ********************************************************************************
6 */
7
8 #include "unicode/utypes.h"
9
10 #if !UCONFIG_NO_BREAK_ITERATION
11
12 #include "unicode/ubrk.h"
13
14 #include "unicode/brkiter.h"
15 #include "unicode/uloc.h"
16 #include "unicode/ustring.h"
17 #include "unicode/uchriter.h"
18 #include "unicode/rbbi.h"
19 #include "rbbirb.h"
20 #include "uassert.h"
21
22 U_NAMESPACE_USE
23
24 //------------------------------------------------------------------------------
25 //
26 // ubrk_open Create a canned type of break iterator based on type (word, line, etc.)
27 // and locale.
28 //
29 //------------------------------------------------------------------------------
30 U_CAPI UBreakIterator* U_EXPORT2
ubrk_open(UBreakIteratorType type,const char * locale,const UChar * text,int32_t textLength,UErrorCode * status)31 ubrk_open(UBreakIteratorType type,
32 const char *locale,
33 const UChar *text,
34 int32_t textLength,
35 UErrorCode *status)
36 {
37
38 if(U_FAILURE(*status)) return 0;
39
40 BreakIterator *result = 0;
41
42 switch(type) {
43
44 case UBRK_CHARACTER:
45 result = BreakIterator::createCharacterInstance(Locale(locale), *status);
46 break;
47
48 case UBRK_WORD:
49 result = BreakIterator::createWordInstance(Locale(locale), *status);
50 break;
51
52 case UBRK_LINE:
53 result = BreakIterator::createLineInstance(Locale(locale), *status);
54 break;
55
56 case UBRK_SENTENCE:
57 result = BreakIterator::createSentenceInstance(Locale(locale), *status);
58 break;
59
60 case UBRK_TITLE:
61 result = BreakIterator::createTitleInstance(Locale(locale), *status);
62 break;
63
64 default:
65 *status = U_ILLEGAL_ARGUMENT_ERROR;
66 }
67
68 // check for allocation error
69 if (U_FAILURE(*status)) {
70 return 0;
71 }
72 if(result == 0) {
73 *status = U_MEMORY_ALLOCATION_ERROR;
74 return 0;
75 }
76
77
78 UBreakIterator *uBI = (UBreakIterator *)result;
79 if (text != NULL) {
80 ubrk_setText(uBI, text, textLength, status);
81 }
82 return uBI;
83 }
84
85
86
87 //------------------------------------------------------------------------------
88 //
89 // ubrk_openRules open a break iterator from a set of break rules.
90 // Invokes the rule builder.
91 //
92 //------------------------------------------------------------------------------
93 U_CAPI UBreakIterator* U_EXPORT2
ubrk_openRules(const UChar * rules,int32_t rulesLength,const UChar * text,int32_t textLength,UParseError * parseErr,UErrorCode * status)94 ubrk_openRules( const UChar *rules,
95 int32_t rulesLength,
96 const UChar *text,
97 int32_t textLength,
98 UParseError *parseErr,
99 UErrorCode *status) {
100
101 if (status == NULL || U_FAILURE(*status)){
102 return 0;
103 }
104
105 BreakIterator *result = 0;
106 UnicodeString ruleString(rules, rulesLength);
107 result = RBBIRuleBuilder::createRuleBasedBreakIterator(ruleString, parseErr, *status);
108 if(U_FAILURE(*status)) {
109 return 0;
110 }
111
112 UBreakIterator *uBI = (UBreakIterator *)result;
113 if (text != NULL) {
114 ubrk_setText(uBI, text, textLength, status);
115 }
116 return uBI;
117 }
118
119
120
121
122
123 U_CAPI UBreakIterator * U_EXPORT2
ubrk_safeClone(const UBreakIterator * bi,void *,int32_t * pBufferSize,UErrorCode * status)124 ubrk_safeClone(
125 const UBreakIterator *bi,
126 void * /*stackBuffer*/,
127 int32_t *pBufferSize,
128 UErrorCode *status)
129 {
130 if (status == NULL || U_FAILURE(*status)){
131 return NULL;
132 }
133 if (bi == NULL) {
134 *status = U_ILLEGAL_ARGUMENT_ERROR;
135 return NULL;
136 }
137 if (pBufferSize != NULL) {
138 int32_t inputSize = *pBufferSize;
139 *pBufferSize = 1;
140 if (inputSize == 0) {
141 return NULL; // preflighting for deprecated functionality
142 }
143 }
144 BreakIterator *newBI = ((BreakIterator *)bi)->clone();
145 if (newBI == NULL) {
146 *status = U_MEMORY_ALLOCATION_ERROR;
147 } else {
148 *status = U_SAFECLONE_ALLOCATED_WARNING;
149 }
150 return (UBreakIterator *)newBI;
151 }
152
153
154
155 U_CAPI void U_EXPORT2
ubrk_close(UBreakIterator * bi)156 ubrk_close(UBreakIterator *bi)
157 {
158 delete (BreakIterator *)bi;
159 }
160
161 U_CAPI void U_EXPORT2
ubrk_setText(UBreakIterator * bi,const UChar * text,int32_t textLength,UErrorCode * status)162 ubrk_setText(UBreakIterator* bi,
163 const UChar* text,
164 int32_t textLength,
165 UErrorCode* status)
166 {
167 UText ut = UTEXT_INITIALIZER;
168 utext_openUChars(&ut, text, textLength, status);
169 ((BreakIterator*)bi)->setText(&ut, *status);
170 // A stack allocated UText wrapping a UChar * string
171 // can be dumped without explicitly closing it.
172 }
173
174
175
176 U_CAPI void U_EXPORT2
ubrk_setUText(UBreakIterator * bi,UText * text,UErrorCode * status)177 ubrk_setUText(UBreakIterator *bi,
178 UText *text,
179 UErrorCode *status)
180 {
181 ((BreakIterator*)bi)->setText(text, *status);
182 }
183
184
185
186
187
188 U_CAPI int32_t U_EXPORT2
ubrk_current(const UBreakIterator * bi)189 ubrk_current(const UBreakIterator *bi)
190 {
191
192 return ((BreakIterator*)bi)->current();
193 }
194
195 U_CAPI int32_t U_EXPORT2
ubrk_next(UBreakIterator * bi)196 ubrk_next(UBreakIterator *bi)
197 {
198
199 return ((BreakIterator*)bi)->next();
200 }
201
202 U_CAPI int32_t U_EXPORT2
ubrk_previous(UBreakIterator * bi)203 ubrk_previous(UBreakIterator *bi)
204 {
205
206 return ((BreakIterator*)bi)->previous();
207 }
208
209 U_CAPI int32_t U_EXPORT2
ubrk_first(UBreakIterator * bi)210 ubrk_first(UBreakIterator *bi)
211 {
212
213 return ((BreakIterator*)bi)->first();
214 }
215
216 U_CAPI int32_t U_EXPORT2
ubrk_last(UBreakIterator * bi)217 ubrk_last(UBreakIterator *bi)
218 {
219
220 return ((BreakIterator*)bi)->last();
221 }
222
223 U_CAPI int32_t U_EXPORT2
ubrk_preceding(UBreakIterator * bi,int32_t offset)224 ubrk_preceding(UBreakIterator *bi,
225 int32_t offset)
226 {
227
228 return ((BreakIterator*)bi)->preceding(offset);
229 }
230
231 U_CAPI int32_t U_EXPORT2
ubrk_following(UBreakIterator * bi,int32_t offset)232 ubrk_following(UBreakIterator *bi,
233 int32_t offset)
234 {
235
236 return ((BreakIterator*)bi)->following(offset);
237 }
238
239 U_CAPI const char* U_EXPORT2
ubrk_getAvailable(int32_t index)240 ubrk_getAvailable(int32_t index)
241 {
242
243 return uloc_getAvailable(index);
244 }
245
246 U_CAPI int32_t U_EXPORT2
ubrk_countAvailable()247 ubrk_countAvailable()
248 {
249
250 return uloc_countAvailable();
251 }
252
253
254 U_CAPI UBool U_EXPORT2
ubrk_isBoundary(UBreakIterator * bi,int32_t offset)255 ubrk_isBoundary(UBreakIterator *bi, int32_t offset)
256 {
257 return ((BreakIterator*)bi)->isBoundary(offset);
258 }
259
260
261 U_CAPI int32_t U_EXPORT2
ubrk_getRuleStatus(UBreakIterator * bi)262 ubrk_getRuleStatus(UBreakIterator *bi)
263 {
264 return ((BreakIterator*)bi)->getRuleStatus();
265 }
266
267 U_CAPI int32_t U_EXPORT2
ubrk_getRuleStatusVec(UBreakIterator * bi,int32_t * fillInVec,int32_t capacity,UErrorCode * status)268 ubrk_getRuleStatusVec(UBreakIterator *bi, int32_t *fillInVec, int32_t capacity, UErrorCode *status)
269 {
270 return ((BreakIterator*)bi)->getRuleStatusVec(fillInVec, capacity, *status);
271 }
272
273
274 U_CAPI const char* U_EXPORT2
ubrk_getLocaleByType(const UBreakIterator * bi,ULocDataLocaleType type,UErrorCode * status)275 ubrk_getLocaleByType(const UBreakIterator *bi,
276 ULocDataLocaleType type,
277 UErrorCode* status)
278 {
279 if (bi == NULL) {
280 if (U_SUCCESS(*status)) {
281 *status = U_ILLEGAL_ARGUMENT_ERROR;
282 }
283 return NULL;
284 }
285 return ((BreakIterator*)bi)->getLocaleID(type, *status);
286 }
287
288
ubrk_refreshUText(UBreakIterator * bi,UText * text,UErrorCode * status)289 void ubrk_refreshUText(UBreakIterator *bi,
290 UText *text,
291 UErrorCode *status)
292 {
293 BreakIterator *bii = reinterpret_cast<BreakIterator *>(bi);
294 bii->refreshInputText(text, *status);
295 }
296
297
298
299 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
300